|
|
@ -36,11 +36,12 @@ do_validate() {
|
|
|
|
do_binarization() {
|
|
|
|
do_binarization() {
|
|
|
|
# Binarize the images
|
|
|
|
# Binarize the images
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
ocrd_olena_binarize_paramters='{"impl": "sauvola-ms-split"}'
|
|
|
|
remove_filegrp OCR-D-IMG-BINPAGE mets.xml
|
|
|
|
remove_filegrp OCR-D-IMG-BINPAGE mets.xml
|
|
|
|
remove_filegrp OCR-D-IMG-BIN mets.xml
|
|
|
|
remove_filegrp OCR-D-IMG-BIN mets.xml
|
|
|
|
ocrd-olena-binarize -l $LOG_LEVEL \
|
|
|
|
ocrd-olena-binarize -l $LOG_LEVEL \
|
|
|
|
-m mets.xml -I OCR-D-IMG -O OCR-D-IMG-BINPAGE \
|
|
|
|
-m mets.xml -I OCR-D-IMG -O OCR-D-IMG-BINPAGE \
|
|
|
|
-p '{"impl": "sauvola-ms-split"}'
|
|
|
|
-p "$ocrd_olena_binarize_paramters"
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
do_linesegmentation_tesserocr() {
|
|
|
|
do_linesegmentation_tesserocr() {
|
|
|
@ -64,11 +65,12 @@ do_linesegmentation_sbb() {
|
|
|
|
# Segment the lines in the images
|
|
|
|
# Segment the lines in the images
|
|
|
|
# TODO: Check that this works with the RGB images
|
|
|
|
# TODO: Check that this works with the RGB images
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
ocrd_sbb_textline_detector_parameters='{"model": "/var/lib/textline_detection"}'
|
|
|
|
remove_filegrp OCR-D-SEG-REGION mets.xml
|
|
|
|
remove_filegrp OCR-D-SEG-REGION mets.xml
|
|
|
|
remove_filegrp OCR-D-SEG-LINE mets.xml
|
|
|
|
remove_filegrp OCR-D-SEG-LINE mets.xml
|
|
|
|
ocrd-sbb-textline-detector -l $LOG_LEVEL \
|
|
|
|
ocrd-sbb-textline-detector -l $LOG_LEVEL \
|
|
|
|
-m mets.xml -I OCR-D-IMG-BINPAGE -O OCR-D-SEG-LINE \
|
|
|
|
-m mets.xml -I OCR-D-IMG-BINPAGE -O OCR-D-SEG-LINE \
|
|
|
|
-p '{"model": "/var/lib/textline_detection"}'
|
|
|
|
-p "$ocrd_sbb_textline_detector_parameters"
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
do_ocr() {
|
|
|
|
do_ocr() {
|
|
|
|